Acoustic Emission-based Assessment of Reinforcement Corrosion: Damage Monitoring and Bond Modelling of Corrosion Effects in Reinforced Concrete

Reinforcement corrosion is believed to be the major deterioration phenomenon in reinforced concrete (RC) structures, causing large direct and indirect financial costs. While tools for modelling and design of new RC structures are mature, methods for modelling and assessment of the remaining structural capacity of existing RC structures are much less developed. A prerequisite for the assessment of these structures is the development of reliable techniques for experimental and in-situ damage level rating, as well as the development of accurate models that can describe the complex structural effects of reinforcement corrosion.

This thesis covers both monitoring and modelling of corrosion-induced damage in RC. Three objectives are set. The first objective is the development and upscaling of passive and active acoustic emission (AE) monitoring protocols for the detection, localisation, quantification, and characterisation of corrosion-induced damage. The second objective is to obtain reliable data sets to investigate the effect of corrosion and to develop reliable models. The third and final objective is the development of a finite element (FE) model to determine bond-slip relations that describe the reinforcement-concrete interface behaviour.

A first part of the thesis focusses on the upscaling of the AE technique. Therefore, an extensive experimental program has been set up to develop AE protocols for corrosion-induced damage in RC. The protocols were applied on three sample sizes: small mortar cylinders (scale 1), RC prisms (scale 2), and RC beams (scale 3). It is found that the AE technique is able to detect corrosion in RC. However, dedicated filtering is necessary in order to reliably localise and characterise AE events. Therefore, a post-processing protocol is developed which significantly improves the localisation results. It is also found that concrete cracking can be identified from cumulative AE energy curves and the peak and centre frequency of the AE signals. Moreover, the AE technique is able to determine the onset of concrete cracking earlier than a visual inspection. Unfortunately, the AE technique does not allow to accurately quantify the amount of mass loss of the rebar due to corrosion. A clustering algorithm is developed to characterise AE sources. The results are compared with micro-CT images (scale 1) and dummy samples (scale 2 and scale 3) to assign possible AE damage sources to the different clusters such as corrosion and concrete cracking.

In a second part, mechanical tests are performed which consist of pull-out tests on RC prisms, and three-point bending tests on RC beams. 

Pull-out tests are performed to investigate the difference in rebar type and corrosion level, as well as the effect of confinement on the pull-out behaviour. A major contribution is the establishment of corrosion level-bond relations and crack width-bond relations, almost non-existent in the literature for smooth rebars. It is found that AE curves enable distinguishing between the type of rebar and the corrosion level. Although the localisation result is significantly improved by application of the developed post-processing protocol, accurate localisation of the crack growth during the pull-out test is shown to be more difficult to localise correctly when the sample is heavily damaged. 

Three-point bending tests are performed on concrete beams reinforced with a ribbed rebar. For high corrosion levels, the failure load decreases and the failure mode shifts from bending to debonding. In this case, no clear correlation is observed between the difference in corrosion level and cumulative AE events and cumulative AE energy. The application of the clustering algorithm, however, allows to distinguish between the corrosion levels.

In a third part, an FE bond model is developed which is calibrated and validated on the previously described pull-out tests. The aim of the model is to obtain bond-slip relations that can be implemented in FE models to describe the structural capacity of corroded RC beams. It is found that incorporating the corrosion accomodation region and the flow of corrosion products into corrosion-induced cracks is important to determine the pressure build-up of corrosion products. A two-phased approach is developed, containing a crack model to step-wise calculate the crack volume, and a bond model to simulate the pull-out behaviour. Every modelling step is considered carefully without overcomplicating the modelling approach. The results of the model are in good agreement with the experimental data for all corrosion levels under investigation.